This work aims to revise the Lanzano, Luzi, Pacor, et al. [2] ground-motion model for shallow crustal earthquakes in Italy (hereinafter, ITA18), calibrated in the magnitude range 4.0–8.0 using strong-motion data recorded up to the recent 2016–2017 Central Italy sequence, and including vibration periods up to 10 s. The improvement of ITA18 is needed because of the possible im- provement of the seismic hazard analysis given by the inclusion of spatial correlation in the context of mixed-effects models. In particular, from the inherent characteristic of seismic data to be spatially defined over the ter- ritory, comes the necessity to develop a method that includes the spatially dependence of the data for the analysis of Intensity Measures of earthquakes events. The approach consists in combining pre-existing techniques of linear mixed-effects modeling, used also by Lanzano, Luzi, Pacor, et al. [2], with the inclusion of spatial dependence in the residuals, by applying an algorithm that resembles the universal kriging. The analysis results in a linear mixed- effects model, with one random effect included all the time, that takes into account also the spatial dependence of the data and results to be better than a simple linear mixed-effects models. So, even if it complicates the model, the insertion of the spatial dependence is worth to be included in the model formulation.
Questo lavoro ha lo scopo di rivedere il modello di movimento del suolo di Lan- zano, Luzi, Pacor, et al. [2] per i terremoti di superficie in Italia (di seguito chiamato ITA18). Questo modello `e calibrato nell’intervallo di magnitudo 4.0-8.0, utilizzando dati di strong-motion registrati fino alla recente sequenza sismica del Centro Italia del periodo 2016-2017, e sono stati inclusi periodi di vibrazione fino a 10 s. L’ampliamento del modello di ITA18 `e necessario per il possibile miglioramento dell’analisi di rischio sismico dato dall’inclusione della correlazione spaziale nel contesto dei modelli a effetti misti. In par- ticolare, dato il carattere intrinseco dei dati sismici di essere spazialmente definiti sul territorio, nasce la necessita` di sviluppare un metodo che includa la dipendenza spaziale dei dati per l’analisi delle misure di intensit`a degli eventi sismici. L’approccio consiste nel combinare tecniche pre-esistenti di modellazione lineare a effetti misti, utilizzate anche da Lanzano, Luzi, Pacor, et al. [2], con l’inclusione della dipendenza spaziale nei residui, applicando un algoritmo simile allo universal kriging. L’analisi risulta in un modello lineare a effetti misti, con un unico effetto randomico incluso, che tiene conto an- che della dipendenza spaziale dei dati e risulta essere migliore di un semplice modello lineare a effetti misti. Di conseguenza, anche se il modello risulta maggiormente complicato, vale la pena includere la dipendenza spaziale nella formulazione del modello.
Spatial mixed-effects models for seismic ground motion estimation : a case study in Italy
PRINCIPE, FEDERICA
2020/2021
Abstract
This work aims to revise the Lanzano, Luzi, Pacor, et al. [2] ground-motion model for shallow crustal earthquakes in Italy (hereinafter, ITA18), calibrated in the magnitude range 4.0–8.0 using strong-motion data recorded up to the recent 2016–2017 Central Italy sequence, and including vibration periods up to 10 s. The improvement of ITA18 is needed because of the possible im- provement of the seismic hazard analysis given by the inclusion of spatial correlation in the context of mixed-effects models. In particular, from the inherent characteristic of seismic data to be spatially defined over the ter- ritory, comes the necessity to develop a method that includes the spatially dependence of the data for the analysis of Intensity Measures of earthquakes events. The approach consists in combining pre-existing techniques of linear mixed-effects modeling, used also by Lanzano, Luzi, Pacor, et al. [2], with the inclusion of spatial dependence in the residuals, by applying an algorithm that resembles the universal kriging. The analysis results in a linear mixed- effects model, with one random effect included all the time, that takes into account also the spatial dependence of the data and results to be better than a simple linear mixed-effects models. So, even if it complicates the model, the insertion of the spatial dependence is worth to be included in the model formulation.File | Dimensione | Formato | |
---|---|---|---|
TESI PRINCIPE FEDERICA.pdf
accessibile in internet solo dagli utenti autorizzati
Dimensione
19.61 MB
Formato
Adobe PDF
|
19.61 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/187621